Deep neural networks can be used to diagnose and detect plant diseases, helping to avoid the plant health-related crop production losses ranging from to % annually. However, the data collection and annotation required to achieve high accuracies can be expensive and sometimes very di cult to obtain in specific use-cases. To this end, this work proposes a synthetic data generation pipeline based on generative adversarial networks (GANs), allowing users to artificially generate images to augment their small datasets through its web interface. The image-generation pipeline is tested on a home-collected dataset of whitefly pests, Bemisia tabaci, on di erent crop types. The data augmentation is shown to improve the performance of lightweight object detection models when the dataset size is increased from to images, seeing a jump in recall at . IoU from . to . %, and an increase in the average IoU from . to . %, without the use of GANs. When GANs are used to increase the number of source object masks and further diversify the dataset, there is an additional . and . % increase in recall and average IoU, respectively. The authenticity of the generated data is also validated by human reviewers, who reviewed the GANs generated data and scored an average of % in distinguishing fake from real insects for low-resolutions sets, and % for high-resolution sets.

There has been an amplified focus on and benefit from the adoption of artificial intelligence (AI) in medical imaging applications. However, deep learning approaches involve training with massive amounts of annotated data in order to guarantee generalization and achieve high accuracies. Gathering and annotating large sets of training images require expertise which is both expensive and time-consuming, especially in the medical field. Furthermore, in health care systems where mistakes can have catastrophic consequences, there is a general mistrust in the black-box aspect of AI models. In this work, we focus on improving the performance of medical imaging applications when limited data is available while focusing on the interpretability aspect of the proposed AI model. This is achieved by employing a novel transfer learning framework, progressive transfer learning, an automated annotation technique and a correlation analysis experiment on the learned representations. Progressive transfer learning helps jump-start the training of deep neural networks while improving the performance by gradually transferring knowledge from two source tasks into the target task. It is empirically tested on the wrist fracture detection application by first training a general radiology network RadiNet and using its weights to initialize RadiNetwrist , that is trained on wrist images to detect fractures. Experiments show that RadiNetwrist achieves an accuracy of 87% and an AUC ROC of 94% as opposed to 83% and 92% when it is pre-trained on the ImageNet dataset. This improvement in performance is investigated within an explainable AI framework. More concretely, the learned deep representations of RadiNetwrist are compared to those learned by the baseline model by conducting a correlation analysis experiment. The results show that, when transfer learning is gradually applied, some features are learned earlier in the network. Moreover, the deep layers in the progressive transfer learning framework are shown to encode features that are not encountered when traditional transfer learning techniques are applied. In addition to the empirical results, a clinical study is conducted and the performance of RadiNetwrist is compared to that of an expert radiologist. We found that RadiNetwrist exhibited similar performance to that of radiologists with more than 20 years of experience. This motivates follow-up research to train on more data to feasibly surpass radiologists’ performance, and investigate the interpretability of AI models in the healthcare domain where the decision-making process needs to be credible and transparent.

No abstract